Journal of Educational Psychology © 2012 American Psychological Association

2012, Vol. 104, No. 4, 922–931 0022-0663/12/$12.00 DOI: 10.1037/a0028661

Selective Benefits of Question Self-Generation and Answering for

Remembering Expository Text
Julie M. Bugg and Mark A. McDaniel
Washington University in St. Louis

The present study examined possible memory and metacomprehension benefits of using a combined
question self-generation and answering technique, relative to rereading, as a study strategy for expository
passages. In the 2 question self-generation and answering conditions (detail or conceptual questions),
participants were prompted on how to generate questions of a particular type (detail or conceptual) and
given practice and feedback prior to reading and studying 4 experimental passages. Participants then
made judgments of learning for detailed and conceptual information from the passages, following which
a cued-recall test with detail and conceptual questions was administered. The self-generation and
answering of conceptual questions yielded a significant benefit to memory performance for conceptual
but not detailed test questions, relative to a rereading condition, whereas the self-generation and
answering of detail questions provided no benefit. A similar pattern was found for metacomprehension
as assessed by calibration, but not relative monitoring accuracy. The selective memory benefit observed
here is consistent with theoretical frameworks that emphasize the importance of transfer- and material-
appropriate processing in modulating the benefits of using question self-generation and answering as a
study strategy.

Keywords: study strategies, self-generated questions, memory, metacomprehension

In educational settings, reading assignments are a ubiquitous Accordingly, much research in education and psychology has
aspect of instruction, particularly at the college level (e.g., reading been directed at identifying techniques to stimulate learners to
assignments are a core component of college syllabi). Theorists more actively and effectively process text materials. In this study,
and educators alike would agree that effective learning requires we examine one such technique, that of requiring readers to
that students engage in active, elaborative processing while read- self-generate and answer questions about the content in the text.
ing (see, e.g., McNamara, 2004; Rosenshine, Meister, & Chapman, This is a technique that Mayer (2003) has advocated as a means for
1996) and that students accurately monitor their comprehension so making the relatively passive activity of reading a more active
as to effectively calibrate their degree of learning (Dunlosky & learning experience. The idea is that self-generating and answering
Metcalfe, 2009). Unfortunately, research suggests that some read- questions encourages readers to elaborate and consider the con-
ers may not ordinarily engage processing strategies that support tents of the text more fully than they otherwise would, leading to
optimal learning. For instance, according to some theories, char- better learning of the text material (for a review, see Wong, 1985).
acteristically some readers are “lazy,” creating only the minimal
Experimental investigations, however, have not uniformly sup-
representational structure necessary to comprehend the text (see
ported this claim, likely in part because as reviewed in Wong and
Fletcher & Bloom, 1988) and failing to pause to construct infer-
Rosenshine et al. (1996), a wide range of methods have been used
ences necessary for complete understanding of technical text
to stimulate self-generation of questions (and additionally, answer-
(Noordman, Vonk, & Kempff, 1992). Furthermore, some readers,
even at the college level, appear to be relatively inaccurate at ing of these questions, in some studies).
gauging how much they understand and will remember about the These reviews suggest generally that methods incorporating
content of an expository passage (see Glenberg & Epstein, 1985; procedural prompts (i.e., training/guiding readers in self-
Maki & Berry, 1984). generating questions) are more effective in improving comprehen-
sion and retention of target content than are methods that do not
incorporate procedural prompts (e.g., Davey & McBride, 1986;
This article was published Online First May 28, 2012. though see Foos, Mora, & Tkacz, 1994, for evidence that prompts
Julie M. Bugg and Mark A. McDaniel, Department of Psychology, are not necessary to produce benefits of question self-generation
Washington University in St. Louis. and answering). But even within the domain of procedural
This research was supported by collaborative activity James S. McDon- prompts, a number of different types of prompts have been used,
nell Foundation Grant 220020166. The authors are grateful to Jivata Raja, and the results in the literature vary somewhat unsystematically
Sneha Thakur, and Jessye Brick for assistance with data collection and
across prompt type. This state of affairs prompted Wong (1985) to
scoring.
Correspondence concerning this article should be addressed to Julie M.
conclude that only by identifying those psychological processes
Bugg, Washington University in St. Louis, Department of Psychology, that are elicited by specific types of questions (prompts) can a clear
Campus Box 1125, One Brookings Drive, St. Louis, MO 63130. E-mail: explanation of what mediates benefits of self-question generation
[email protected] be attained.

922
 GENERATION AND ANSWERING OF QUESTIONS 923

As a step in this direction, we examined in the present study because it is assumed that accurate metacomprehension effectively
relatively well-defined prompts that are characterized as “question guides subsequent study (i.e., students spend additional time study-
types” (Rosenshine et al., 1996). Raphael and Pearson (1985) ing passages deemed to be not well learned; for a review, see Son
identified three question types, with two of the types in part & Metcalfe, 2000).
motivating the present study. For one type, the answer is found in
a single sentence, and for the second type, the answer requires Theoretical Predictions
integration across two or more sentences. This classification was
developed to assist learners in improving their performance on A priori, several different patterns seemed possible. From the
criterial tests (Raphael & Pearson) and has been extended as perspective that readers will learn more when they are encouraged
prompts to guide readers’ self-generation of questions (Dermody, to more actively process the text, the most straightforward expec-
1988; Labercane & Battle, 1987; Smith, 1977). In the sole pub- tation is that the question self-generation and answering groups
lished study examining the Raphael and Pearson question-type would generally display better memory for the contents of the text
prompts, grade-school children with learning disabilities were than the reread control group. We believed this was not implau-
prompted on how to generate different question types (in addition sible, reasoning that self-generation and answering of questions,
to receiving guidance on other strategies such as summarizing) in even if focused on one particular type of content (detail, concep-
28 sessions of training. The training did not, however, lead to tual), would require at the minimum that readers consciously
significant gains in reading achievement relative to a control evaluate the nature of the information they were considering as the
condition (Labercane & Battle). In the present study, we directly basis for question generation (detail, conceptual). Such elaborative
contrast two question-type prompts similar to those described by processing might benefit retention of both types of information
Raphael and Pearson, and we do so in a single session with college even if the questions that were self-generated and answered were
students. No study has explicitly contrasted the potential benefits limited to one type of information (detail, conceptual). By the
associated with the different question-type prompts of interest same token, metacomprehension accuracy might also be generally
here, but one recent study hints that a single session may be improved by question self-generation. This expectation is based on
sufficient for training college students to self-generate questions findings showing that techniques that stimulate more active pro-
that can improve memory for textually presented information. cessing of text, such as concept mapping and self-explanation,
Weinstein, McDermott, and Roediger (2010) gave college student significantly improve metacomprehension accuracy (e.g., Griffin,
participants “comprehension questions” that other students had Wiley, & Thiede, 2008; Maki, Foley, Kajer, Thompson, & Willert,
previously generated as prompts to guide the kinds of questions 1990; Thiede, Anderson, & Therriault, 2003; Thiede, Griffin,
that the participants should generate and answer during their own Wiley, & Anderson, 2010).
reading. This self-questioning and answering group showed better Alternatively, the transfer-appropriate-processing approach to
memory for the information in the texts than did a group who memory suggests that elaborative processing will benefit memory
reread the texts. Weinstein et al. did not, however, manipulate only to the extent that the processing overlaps with the information
different types of questions that could be self-generated, assess required at test (McDaniel, Friedman, & Bourne, 1978; Morris,
whether the benefits depended on the types of questions generated, Bransford, & Franks, 1977; Roediger, Weldon, & Challis, 1989).
or analyze metacomprehension accuracy; we explore these novel The transfer-appropriate-processing framework can be directly
factors in the present experiment. applied to the present context, generating the following predic-
Following the Raphael and Pearson (1985) framework, we tions. Prompting self-generation and answering of detail questions
prompted different groups of participants to generate and, in ad- should improve memory for details but not conceptual information
dition, answer one of two types of questions to help them study relative to the reread control. By contrast, prompting self-
target passages (i.e., expository texts). For ease of exposition generation and answering of conceptual questions should improve
and to facilitate our theoretical analysis of how these different memory for conceptual information but not details (relative to the
question types might affect memory and metacomprehension, we reread control).
label one type of question a detail question (the answer referred to Recent work has also shown that transfer-appropriate-
a detail or fact that could be found in a single sentence) and the processing dynamics appear to hold for metacomprehension
other type of question a conceptual question (the answer integrated (Thomas & McDaniel, 2007). Thomas and McDaniel (2007) found
thematic information across two or more sentences) (cf. Thomas & that metacomprehension was more accurate for detailed test ques-
McDaniel, 2007). The question self-generation conditions were tions when participants engaged in detailed processing at study
compared with a reread control condition. We considered the (i.e., inserted missing letters in words; see also Maki et al., 1990)
reread condition to be a strong control because it required readers than when they engaged in conceptual processing at study (i.e.,
to spend additional time on the texts, following an initial reading, sorted sentences into a coherent paragraph), with the reverse being
as did the question self-generation groups. To comprehensively found for conceptual test questions. Although these findings are
gauge the mnemonic outcomes of question self-generation and to limited to study strategies that are not educationally authentic, that
provide leverage on the theoretical views developed below, we is, strategies not typically adopted by students or advocated by
administered a final memory test that probed for both details and educators, they do propel the following predictions for the present
conceptual information. In addition, we assessed the accuracy of study. Metacomprehension accuracy for details should be higher
participants’ metacomprehension of the passages—the degree to after self-generation and answering of detail questions, whereas
which the reader can judge his or her own learning of text mate- metacomprehension accuracy for conceptual information should
rials (Dunlosky & Lipko, 2007). Metacomprehension accuracy be higher after self-generation and answering of conceptual ques-
may be an equally important outcome in educational contexts tions, relative to the reread condition.
924 BUGG AND MCDANIEL

There is yet a third, more nuanced and perhaps counterintuitive Design and Procedure
set of predictions based on a contextual framework developed to
anticipate effects of more engaged processing on learning (e.g., A 3 ⫻ 2 mixed subjects design was used, with study strategy as
stimulated by introducing difficulties into the learning environ- a between-subjects factor and type of test question as a within-
ment; see McDaniel & Butler, 2011, for a complete presentation of subjects factor. The three study strategy conditions were reread,
this framework and related evidence). Briefly, this framework self-generation and answering of detailed questions, and self-
assumes that in addition to the transfer appropriateness of process- generation and answering of conceptual questions. During study,
ing, one must consider the processing normally invited by the participants in the reread condition read the entire passage once,
materials themselves (termed material appropriate processing; and then read it a second time. Participants in the detailed and
Einstein, McDaniel, Owen, & Cote, 1990; McDaniel & Einstein, conceptual question generation and answering conditions read the
1989, 2005). Prior research suggests that different types of text entire passage once and then generated either detail questions or
invite encoding and retention of somewhat different types of conceptual questions (and corresponding answers), respectively.
information, with well-structured narratives (such as folktales) During test, detail and conceptual test questions were administered
inviting processing of conceptual information that interrelates to all participants. A detail question was defined as one that
propositions in the text (McDaniel, Hines, Waddill, & Einstein, referred to a detail or fact contained within a single sentence in the
1994) and expository text inviting more focus on details (see passage. A conceptual question referred to the overall gist or a
McDaniel & Einstein, 1989, for a more extended discussion of this major theme and required integration of information across (at
assumption). Furthermore, the effectiveness of any particular study least two) sentences.
strategy is anticipated to depend on the degree to which the study Following informed consent, participants were instructed that
strategy stimulates processing of information that is not normally they would be reading and studying several passages for a later
invited by the particular target passages. test. In the reread condition, participants read the entire passage
For present purposes, the key point regards study strategies that once, informed the experimenter when they were done, and were
are anticipated to be effective for expository text. Because expos- then asked to study the passage by rereading it. Participants in the
itory text invites more focus on details (perhaps because there is an question generation and answering conditions were instructed to
absence of a ready conceptual structure), strategies that stimulate read the entire passage and inform the experimenter when they
processing of relational information (i.e., information that inte- were done. They were then asked to study the passage according
grates across several sentences or propositions) will be comple- to the following prompt:
mentary to the processing already invited by expository passages,
When studying the passages, we would like you to use a unique
and consequently will benefit memory (recall). However, study
strategy in which you imagine that you are a teacher who has assigned
strategies that stimulate processing of details (or what has been the passage to his/her class to read. You, as the teacher, are now
termed “proposition-specific” elaboration) will be relatively re- developing a test to administer to your students to assess their knowl-
dundant with that normally invited by expository text, and conse- edge of information from the passage. For each passage, you will be
quently have little benefit on memory (see Einstein et al., 1990, for asked to generate 3 questions and their corresponding answers.
supporting experiments; see McDaniel & Einstein, 2005, for a
review). Applied to the present question self-generation and an- Participants had access to the passage while generating and an-
swering conditions, this framework anticipates that the self- swering the questions (i.e., while studying). Participants were told
generation and answering of conceptual questions will improve that the questions should assess students’ knowledge of a detail or
memory performance on final conceptual questions relative to the fact from a single sentence in the passage (detail question gener-
reread control. More provocatively, this framework anticipates that ation condition) or knowledge of concepts or information that must
the focus on detailed information presumably prompted by self- be integrated across sentences in the passage (conceptual question
generating and answering detailed questions will produce no sig- generation condition).
nificant improvement for the detail questions on the final test Following the initial set of instructions, an excerpt from a
relative to the reread condition (because the prompted focus on sample passage was shown. Participants in the reread condition
details is assumed to be relatively redundant with that invited by simply read the sample passage, whereas those in the question-
expository texts). With regard to metacomprehension accuracy, we generation conditions read the passage and were shown a sample
thought it possible that the just predicted memory patterns would detail question or a sample conceptual question (depending on the
extend to metacomprehension accuracy. Specifically, self- question-generation condition assigned). The experimenter ex-
generation and answering of conceptual questions might selec- plained why the question was designated as a particular type
tively enhance metacomprehension accuracy for conceptual ques- (detail or conceptual). Next, a practice passage was given. Partic-
tions. ipants in the reread condition read, and then reread the passage.
Those in the question-generation conditions were asked to read the
passage and engage in the study task. For the practice passage,
Method participants were prompted to generate only two (detail or con-
ceptual) questions. Participants wrote the questions and the corre-
Participants sponding answers on a response sheet. The experimenter then
provided the participant with individualized feedback pertaining to
Forty-eight undergraduates participated for course credit or the questions they had generated. They were told whether or not
were paid at a rate of $10/hr. Sixteen participants were randomly the questions were of the type (detail or conceptual) requested and,
assigned to each of the three study strategy conditions. if not, why they were not and how the questions could be changed
 GENERATION AND ANSWERING OF QUESTIONS 925

to conform to the guidelines provided. Finally, the experimenter tica,” with the answer “Crevasses are concealed, and they are
provided two sample questions of the type requested and con- constantly moving.” This answer requires integration of informa-
firmed that the participant understood the study instructions. tion across the following contiguous sentences from the text:
The experimental phase consisted of four passages. One passage
was presented at a time, with the next passage provided by the The glacier is forced to change direction, causing an enormous
experimenter after the participant read and studied the previous build-up in pressure. Many crevasses are formed by this process also.
Crevasses may be a concealed peril for travelers. Crevasses are often
passage. The passages were given in the same order to all partic-
shifted great distances from the place they were created as the glacier
ipants. Participants (except those in the reread condition) recorded
moves. Unconcealed crevasses can easily be avoided, but most are
self-generated questions and answers on a different response sheet hidden by snow bridges. The wind builds snow bridges by packing
for each passage. After participants read and studied all four snow across the opening of the chasm. Eventually, the bridge and the
passages, they were asked to make judgments of learning. Specif- crevasse become completely hidden by drifting snow.
ically, they were asked to judge how well they would remember
the information that was contained in each of the passages using a
Results
scale ranging from 0 (extremely unlikely to remember) to 100
(extremely likely to remember). Participants were encouraged to The alpha level was set at .05. Partial eta squared (␩2p) is
use the full range of the scale when making their judgments. reported as the measure of effect size for all significant effects.
Participants provided separate judgments for each passage, begin-
ning with the first passage they studied and moving in order to the
last. Importantly, for each passage, one judgment was made for Manipulation Checks
details/facts, and a second judgment was made for concepts/ To confirm that participants in the generation and answering
integrative information. Participants in the detail question- conditions adhered to the instructions they were provided, we first
generation and answering condition made the detail/fact judgment calculated the mean percentage of questions that were of the type
first, and then made the concept/integrative information judgment. requested (i.e., detail for the detail question-generation/answering
The reverse was true for participants in the conceptual question- condition and conceptual for the conceptual question-generation/
generation and answering condition. The order of these judgments answering condition). Adherence did not differ between the detail
was counterbalanced for participants in the reread condition. question-generation and answering group (M ⫽ 98%, SE ⫽ 2%)
Following the metacognitive judgments, a cued-recall test was and the conceptual question-generation and answering group (M ⫽
given. The test consisted of 24 questions (three detail and three 93%, SE ⫽ 4%), t(30) ⫽ 1.03, p ⫽ .311. Next, we calculated the
conceptual questions per passage). The six test questions that mean percentage of self-generated questions that were correctly
corresponded to the first passage participants had read and studied answered. Accuracy was perfect (or nearly perfect) for participants
were administered, followed by those corresponding to the second in the detail (M ⫽ 100%, SE ⫽ 0%) and conceptual (M ⫽ 99%,
passage, and so on. Participants were shown the title of the SE ⫽ 1%) question and answering conditions, t(30) ⫽ 1.46, p ⫽
passage, then the six questions that were administered in a random .154. Together, these findings indicate that participants in both
order. Following completion of the cued-recall test, participants groups followed the instructions equally well.
were thanked and debriefed.

Cued Recall
Materials
Two raters scored responses to the cued-recall test questions.
Following Thomas and McDaniel (2007), six expository texts
We used a partial credit scoring procedure whereby participants
from Levy (1981) were used, which were taken from the interme-
were awarded 0, .5, or 1 point per question.1 Interrater reliability
diate level of the Science Research Associates (SRA) reading
was high (r ⫽ .90). The measure of cued-recall performance was
series. The sample passage was an excerpt from the text “Skunks.”
average points per question earned (e.g., a participant who earned
The text “How Autumn Colors are Formed” was used as the
1 point for six of the 12 detail questions and 0 points for the other
practice passage. The experimental passages, in the order shown to
six would have a .5 score for cued-recall performance on the detail
participants, were the texts “Kanchenjunga: A Very Dangerous
questions), and this measure was calculated separately for the 12
Mountain Range”; “Nomads of the Desert”; “The Strange Way of
detail and 12 conceptual questions.
Spiders”; and “The Frozen Continent.” The texts ranged in length
We conducted a 3 (study strategy: reread vs. generation of detail
from 285 to 360 words and were presented in a single-paragraph
questions vs. generation of conceptual questions) ⫻ 2 (type of test
format.
question: detail vs. conceptual) mixed subjects analysis of variance
Three detail and three conceptual questions were generated for
(ANOVA) for cued-recall performance. Significant main effects of
each passage for the cued-recall test. Half of these questions
study strategy, F(2, 45) ⫽ 3.21, MSE ⫽ .039, p ⫽ .050, ␩2p ⫽ .125,
overlapped with those used by Thomas and McDaniel (2007), and
and type of test question, F(1, 45) ⫽ 13.68, MSE ⫽ .016, p ⫽ .001,
half of the questions were newly developed. For “The Frozen
␩2p ⫽ .233, were qualified by a significant interaction between
Continent,” an example of a detail question is: “How many square
miles in size is Antarctica’s great ice cap?” The answer, “six
million” comes from the following single sentence from the text: 1
We report partial credit-based performance because this scoring pro-
“Antarctica’s great ice cap alone is six million square miles in cedure is more common in educational contexts. Note, however, that use of
size.” An example of a conceptual question is: “Give two reasons a strict (i.e., nonpartial credit) scoring procedure yields a highly similar
why it is impossible to create a map of the crevasses in Antarc- pattern of interactive effects for both cued recall and calibration.
926 BUGG AND MCDANIEL

study strategy and type of test question, F(2, 45) ⫽ 3.55, MSE ⫽ each of the question-generation and answering conditions, we
.016, p ⫽ .037, ␩2p ⫽ .136. compared the cued-recall performance of participants with rela-
To test the hypotheses outlined in the introduction, we decom- tively high and low question overlap. For the conceptual question-
posed the 3 ⫻ 2 interaction by analyzing the simple effects of generation condition, conceptual test performance was signifi-
study strategy using one-way ANOVAs and the simple effects of cantly higher for participants (n ⫽ 7) whose questions overlapped
question type using dependent t tests. Cued-recall performance for 33%–50% (50% was the highest) with the final test (M ⫽ .77,
the detail questions did not differ across the three study strategy SE ⫽ .03) than for participants (n ⫽ 9) whose questions had
groups (F ⬍ 1) (see Figure 1). By contrast, cued-recall perfor- relatively low (0%–25%) overlap (M ⫽ .58, SE ⫽ .06), t(14) ⫽
mance for the conceptual questions did vary significantly as a ⫺2.58, p ⫽ .022. For those whose questions had relatively low
function of study strategy, F(2, 45) ⫽ 5.33, MSE ⫽ .064, p ⫽ .008, overlap, cued-recall performance approached that observed in the
␩2p ⫽ .191 (see Figure 1). Bonferroni post hoc comparisons indi- reread condition (M ⫽ .50, SE ⫽ .05). For the detailed question-
cated that a significant benefit was observed for the group that generation condition, there was minimal effect of the degree of
generated and answered conceptual questions (M ⫽ .66, SE ⫽ .04) overlap for detail test performance. Participants (n ⫽ 7) whose
relative to the group that generated and answered detail questions questions had relatively high (33%–50%; 42% was the highest)
(M ⫽ .47, SE ⫽ .04, p ⫽ .012) or the group that reread (M ⫽ .50, overlap performed similarly (M ⫽ .44, SE ⫽ .06) to participants
SE ⫽ .05, p ⫽ .043). (n ⫽ 9) whose questions had relatively low (0%–25%) overlap
Comparisons of cued-recall performance on detail and concep- (M ⫽ .40, SE ⫽ .04), t(14) ⫽ ⫺0.53, p ⫽ .603.
tual questions for each of the three groups revealed the following
pattern of results. Cued-recall performance did not differ on the Judgments of Learning
detail and conceptual questions for either the group that generated
and answered detail questions, t(15) ⫽ ⫺1.12, p ⫽ .283, or the To examine whether judgments of learning varied as a function
group that reread, t(15) ⫽ ⫺0.816, p ⫽ .427. However, the group of study strategy and type of test question, we performed a 3 (study
who generated and answered conceptual questions performed sig- strategy) ⫻ 2 (type of test question) mixed ANOVA. The main
nificantly better on the conceptual questions (M ⫽ .66, SE ⫽ .04) effects of study strategy and question type were not significant, but
than on the detail questions (M ⫽ .47, SE ⫽ .04), t(15) ⫽ ⫺5.86, the interaction of these two factors was significant, F(2, 45) ⫽
p ⬍ .001. 4.68, MSE ⫽ .007, p ⫽ .014, ␩2p ⫽ .172 (see Figure 1 for means).
To decompose the interaction, we first analyzed the simple main
Question Overlap and Cued Recall effects of study strategy using one-way ANOVAs. For the detailed
question type, judgments of learning did differ according to study
We examined the percentage of questions participants in each strategy, F(2, 45) ⫽ 3.62, MSE ⫽ .014, p ⫽ .035, ␩2p ⫽ .139.
group generated that overlapped with the questions administered Bonferroni post hoc comparisons indicated that judgments of
on the cued-recall test, as this has been shown to influence final learning were significantly higher for participants who generated
test performance (e.g., Frase & Schwartz, 1975). Overlap was and answered conceptual (M ⫽ .76, SE ⫽ .03) as compared with
similar for the group who generated detail questions (M ⫽ 26%, detailed (M ⫽ .65, SE ⫽ .03) test questions (p ⫽ .038). For the
SE ⫽ 2%) and the group who generated conceptual questions conceptual question type, the effect of study strategy was not
(M ⫽ 28%, SE ⫽ 3%), t(30) ⫽ ⫺0.26, p ⫽ .799. To examine significant, F(2, 45) ⫽ 1.12, p ⫽ .335.
whether overlap was related to cued-recall performance within We then analyzed the simple main effects of question type using
dependent t tests. For the group who generated and answered detail
questions, judgments of learning were nominally but not signifi-
cantly higher for the conceptual (M ⫽ .70, SE ⫽ .03) as compared
with detailed (M ⫽ .65, SE ⫽ .04) test questions, t(15) ⫽ ⫺1.64,
p ⫽ .123. By contrast, for the group who generated and answered
conceptual questions, judgments of learning were lower for the
conceptual (M ⫽ .69, SE ⫽ .03) as compared with detail (M ⫽ .76,
SE ⫽ .02) test questions, t(15) ⫽ 3.20, p ⫽ .006. For the reread
group, judgments of learning were equivalent for conceptual
(M ⫽ .75, SE ⫽ .03) and detailed (M ⫽ .73, SE ⫽ .02) test
questions (t ⬍ 1).

Calibration
To examine metacomprehension accuracy, we derived a mea-
sure of calibration by subtracting actual cued-recall performance
from predicted cued-recall performance (i.e., judgments of learn-
ing) (Dunlosky & Metcalfe, 2009). Positive values for calibration
Figure 1. Actual and predicted cued-recall performance (i.e., judgments reflect overconfidence, whereas negative values reflect undercon-
of learning) for detail and conceptual test questions as a function of study fidence. We conducted a 3 (study strategy) ⫻ 2 (type of test
strategy. Calibration is the difference between actual and predicted perfor- question) mixed ANOVA for these calibration values. A main
mance. Error bars represent standard error. effect of question type, F(1, 45) ⫽ 11.13, MSE ⫽ .021, p ⫽ .002,
 GENERATION AND ANSWERING OF QUESTIONS 927

␩2p ⫽ .198, was qualified by a significant Study Strategy ⫻ Type (a question requiring inference or involving a macrostatement)
of Test Question interaction, F(2, 45) ⫽ 8.12, MSE ⫽ .021, p ⫽ question.3 For example, a low-quality question pertaining to the
.001, ␩2p ⫽ .265. excerpt on crevasses from the “Frozen Continent” text (see the
To decompose this interaction, we conducted the analyses of Method section) is “Why are crevasses in the ice formed and often
simple effects as described above. Mirroring the patterns obtained hidden?” whereas a high-quality question is “Why are crevasses
for cued-recall performance, the three study strategy groups did dangerous formations for Antarctic travelers?” We awarded 1
not differ in calibration for the detail test questions (F ⬍ 1), but did point for each low-quality question and 2 points for each high-
differ in calibration for the conceptual test questions, F(2, 45) ⫽ quality question such that a participant minimally earned 12 points
8.64, MSE ⫽ .056, p ⫽ .001, ␩2p ⫽ .277 (see Figure 1). Bonferroni and maximally earned 24 points. On average, the group who
post hoc comparisons indicated that participants who generated generated conceptual questions earned 16.5 points (SE ⫽ .45),
and answered conceptual questions at study had significantly bet- whereas the group who generated detail questions earned 12.3
ter calibration for the conceptual test questions than participants in points (SE ⫽ .15), a statistically significant difference, t(30) ⫽
the detail question-generation group (p ⫽ .004) or reread group ⫺8.71, p ⬍ .001. As expected, there was little variation within the
(p ⫽ .002); the latter two groups did not differ (p ⬎ .999). detail question-generation group (range ⫽ 12–14). However, there
Turning to the simple effects of question type, calibration was was more variation within the conceptual question-generation
equivalent for detailed and conceptual test questions for both the group (range ⫽ 14 –20), suggesting that some participants pro-
group who generated and answered detail questions, t(15) ⫽ 0.025, duced primarily questions that required a verbatim restatement of
p ⫽ .980, and the group who reread, t(15) ⫽ 0.425, p ⫽ .677. In information, whereas others produced a mixture, including some
contrast, the group who generated and answered conceptual ques- questions that required inferencing or involved macrostatements.
tions had better calibration on the conceptual test questions (M ⫽ Consequently, we thought it informative to examine the correla-
.03, SE ⫽ .04) than on the detailed test questions (M ⫽ .29, SE ⫽ tions within the conceptual question-generation group between
.04), t(15) ⫽ 6.62, p ⬍ .001. In fact, for the conceptual test question quality, memory (cued-recall performance), and meta-
questions, participants who generated and answered conceptual comprehension (calibration). There were no significant correla-
questions had calibration scores that did not differ from zero, tions between quality and memory, or quality and metacompre-
t(15) ⫽ 0.61, p ⫽ .55, suggesting a very high degree of metacom- hension (largest r ⫽ .17, p ⫽ .53).
prehension accuracy. Calibration did differ significantly from zero
for all other combinations of question type and study strategy (ts ⬎ Discussion
5.36, ps ⬍ .001).
In the present study, we examined the effects of a combined
question self-generation and answering technique on memory and
Relative Monitoring Accuracy metacomprehension outcomes. An overarching view has been that
A second measure of metacomprehension accuracy is relative self-generation of questions produces more active reading, includ-
monitoring accuracy. Following Griffin et al. (2008), we computed ing a greater focus of attention on and elaboration of content and
Pearson correlations between each individual’s cued-recall perfor- greater self-awareness of the degree of comprehension (Rosen-
mance and their predicted performance across the four texts,2 save shine et al., 1996), thereby leading to increased comprehension,
for two participants in the detail question-generation group who memory, and metacomprehension. Rosenshine et al.’s review
produced no variability in predicted performance across the texts. highlighted procedural prompts to guide question self-generation
As shown in Table 1, the average correlations were very weak for as an important factor influencing the attainment of such benefits.
both detailed and conceptual test performance, with none differing However, evidence on the effectiveness of question self-generation
from zero (ps ⬎ .18). when accompanied by a procedural prompt is sparse with college
students, with some of the primary reports examining comprehen-
Analysis of the Quality of Generated Questions sion of lectures (King, 1989, 1992) and not texts. An exception is
a recent study by Weinstein et al. (2010); like Weinstein et al., we
Finally, we examined the quality of the generated questions by used a self-generation and answering condition and found support
categorizing each as either a low-quality (a question requiring a for the overarching view. Importantly, however, in the present
verbatim restatement of information from the text) or high-quality study, support was qualified by a consideration of (a) the particular
types of questions and answers prompted for generation, a factor
not considered in previous studies in which the effects of question
Table 1
generation on memory and metacomprehension of text have been
Average Relative Monitoring Accuracy (Standard Error in
examined and (b) the particular types of information targeted in the
Parentheses) as a Function of Study Strategy and Type of
test questions. Specifically, we found that generation and answer-
Test Question
ing of conceptual but not detailed questions benefited performance
Study strategy
2
Detail Conceptual The pattern of results was identical when gamma correlations were
question generation question generation computed.
Type of test Reread and answering and answering 3
We thank several anonymous reviewers for suggesting that we analyze
the quality of the generated questions along these dimensions and examine
Detail 0.20 (.14) ⫺0.05 (.15) ⫺0.07 (.18) possible relationships between question quality, memory, and metacom-
Conceptual 0.04 (.15) 0.10 (.14) ⫺0.16 (.13)
prehension within the conceptual question-generation condition.
928 BUGG AND MCDANIEL

on the final memory test and appeared to improve calibration types of texts used here). By contrast, the processing stimulated by
relative to rereading and that these benefits were specific to con- the generation and answering of conceptual questions is not re-
ceptual test questions. Importantly, this finding cannot be dundant with that invited by the expository texts and accordingly,
explained by differential adherence to question-generation instruc- as we observed, enhances memory for this type of information.
tions across conditions, differential accuracy in answering self- It is worth emphasizing that the average overlap in content targeted
generated questions, or differential degrees of overlap between the by the self-generated conceptual questions and the criterial test con-
questions generated by participants and those that appeared on the ceptual questions was not particularly high (28%), suggesting the
final test, as these measures did not differ between the question- pedagogical utility of stimulating learners to attend to levels of infor-
generation and answering groups. mation that expository texts (at least for the corpus used here) do not
This pattern suggests limits to overly general claims regarding ordinarily invite learners to encode well. Even so, one would expect
the benefits of self-generating and answering questions as a study that the conceptual question prompts would have increasing benefits
strategy and points to a more refined understanding of the effects to the degree that learners could generate questions overlapping with
of self-generation and answering of questions on memory and the criterial test. This expectation was borne out in the present exper-
metacomprehension. Regarding memory, of note are two promi- iment: Conceptual test performance rose to .77 for participants whose
nent components of our findings. The first is that generating and generated conceptual questions overlapped relatively well with the
answering conceptual questions did not produce benefits to per- final test, compared with .58 for participants with relatively low
formance on the detailed questions. This finding counters the overlap.4 Note that this pattern corresponds to the transfer-
possibility outlined in the introduction that generating and answer- appropriate-processing interpretation discussed above; for those par-
ing certain types of questions, such as conceptual questions, would ticipants who generated conceptual questions that did not overlap with
prompt learners to sift through both detail and conceptual (the- the final test (i.e., relatively low overlap in processing of question
matic) information, thereby potentially elaborating both kinds of generation and final test), cued-recall performance approached that
information. Instead, the effect of generating and answering con- observed in the reread condition (.50).
ceptual questions displayed a transfer-appropriate pattern such that For the detailed question-generation and answering condition,
the processing engaged when generating conceptual questions there was minimal effect of the degree of overlap for detail test
benefited performance on final test questions that targeted a sim- performance. This pattern reinforces the conclusion that minimal
ilar level of information (i.e., conceptual). This conclusion might benefits may be associated with self-generation and answering of
be further refined in light of the analysis of the quality of the questions focusing on levels of information that are redundant with
generated questions. As might be expected, the conceptual that afforded by the text (during normal reading). Of course, it
question-generation and answering group generated higher quality remains possible that if learners could somehow generate and
questions (in terms of requiring inferencing and construction of answer questions that targeted tested details with a high degree of
macrostatements) than the detail question-generation and answer- overlap, then self-generation and answering of detailed questions
ing group. However, within the conceptual question-generation could increase learning and memory from expository text (cf. Maki
and answering group, generating higher quality conceptual ques- et al., 1990; Thomas & McDaniel, 2007, for related findings).
tions was not associated with higher recall levels. Accordingly, it For metacomprehension accuracy, as assessed by calibration,
may be that the critical conceptual processing feature for enhanc- we found a similar pattern to that which was observed for memory
ing performance on the conceptual test questions was attempting to performance. Specifically, a selective benefit was obtained for
generate and answer questions that required integration of infor- metacomprehension accuracy on conceptual test questions follow-
mation across sentences, as specified in the instructions. ing the self-generation and answering of conceptual questions.
The second component is that the just-mentioned transfer- This benefit reflected that the typical pattern of overconfidence in
appropriate-processing effect was observed exclusively for the metacomprehension, observed in all other conditions in the present
generation and answering of conceptual questions. That is, a study, was eliminated, and high levels of metacomprehension
benefit to cued-recall performance on detailed questions was not accuracy (i.e., calibration) were obtained when the type of self-
observed for participants who generated and answered detailed generated questions both matched the criterial final test processing
questions at study. This potentially curious finding is readily and were appropriate for (i.e., not redundant with) the materials.
interpreted within the material-appropriate-processing (MAP) To the extent that metacomprehension is used to guide subsequent
framework outlined in the introduction. The MAP framework study, elimination of overconfidence is a finding of practical
suggests that different types of text typically invite readers to importance as it anticipates more appropriate decisions regarding
extract particular types of information (Einstein et al., 1990;
McDaniel, Einstein, Dunay, & Cobb, 1986). For present purposes,
4
the key assumption of this framework is that individuals tend to The fact that performance was at 77% for the highest levels of overlap
focus on details when reading expository texts (presumably be- may counter practical concerns regarding somewhat modest overall abso-
cause there is no obvious organizational structure within which to lute performance levels (both for the reread and the question-generation
integrate the content; see McDaniel & Einstein, 1989, for ampli- groups). That is, in an educational setting, the overall performance levels
could be viewed as unsatisfactory. For purposes of experimental design, we
fication, and Einstein et al., 1990, for supporting findings with
implemented conditions (e.g., limited exposure to the material, use of
some of the texts used in the present study). Applied to the present materials for which participants had little background) to obtain perfor-
pattern, the framework suggests that processing stimulated by the mance levels that would allow sensitivity to differences among experimen-
generation and answering of detailed questions is redundant with tal conditions. Notably, even under these conditions, participants who
that invited by the materials themselves, and therefore detailed tended to generate conceptual questions that overlapped with those on the
questions do not provide benefits relative to reading alone (for the final test showed relatively high performance (77%).
 GENERATION AND ANSWERING OF QUESTIONS 929

study (cf. Thomas & McDaniel, 2007). There is, however, reason in improving memory and metacomprehension accuracy for this
to be cautious in interpreting this finding; the enhanced calibration genre of text (see Einstein et al., 1990). Admittedly, such texts
appears to be more reflective of fluctuations in cued-recall perfor- (e.g., folktales) are rarely found in educational settings. Still, in
mance than it is of enhanced metacognitive monitoring (see, e.g., educational settings, learners with expertise in a content area might
Connor, Dunlosky, & Hertzog, 1997; Griffin, Jee, & Wiley, 2009). easily organize expository texts from that content area (cf.
That is, for the group that generated and answered conceptual McNamara, Kintsch, Butler-Songer, & Kintsch, 1996) and conse-
questions, cued-recall performance increased for conceptual ques- quently could benefit more from detailed question generation than
tions. As such, the generally high judgment of learning that this conceptual question generation. A test of these predictions, in
group assigned to conceptual questions approached actual cued- concert with the present data, would fully inform the material
recall performance for this question type. appropriate processing interpretation of the present patterns.
As for relative monitoring accuracy, the correlations between A more straightforward potential implication of the present
predicted performance and actual performance for the four pas- findings concerns limitations to the value of self-generation and
sages were weak, with none differing from zero. It is possible that answering of questions as a study strategy in educational settings.
the low levels of relative monitoring accuracy in the present study Self-generation and answering of questions takes additional time
may relate to the small number of passages. However, Griffin et al. over rereading (e.g., Weinstein et al., 2010). Consequently, unless
(2008) used only four passages but found that relative monitoring educators are alert to the demands of the criterial test and process-
accuracy was significant (.63) for a different type of active- ing afforded by materials themselves, encouraging students to
processing strategy (self-explanation). A potentially important dif- engage in question generation and answering could prove to be
ference between the present study and that of Griffin et al. is that labor in vain on the student’s part. Specifically, for educational
they obtained judgments immediately after participants read each settings in which students are studying expository text, our find-
text (see also Thomas & McDaniel, 2007), whereas we used a ings suggest that prompting self-question generation and answer-
delayed procedure for collecting metacomprehension judgments. ing may not be optimal when criterial tests focus on details or
Some early studies suggested that relative monitoring accuracy when students are prompted to generate detail questions. We offer
differs from zero only when judgments and tests are immediate these implications cautiously because learner ability, prior knowl-
(Maki, 1998; see also Glenberg & Epstein, 1985, Experiment 2). edge, or even length of the texts (i.e., we used relatively short
However, more recent work has shown that delayed judgment passages) could further qualify the present patterns.
tasks can improve relative monitoring accuracy. For example, it is In closing, we refer to Rosenshine et al.’s (1996) lament “at
improved when the study task (e.g., keywording, summarizing, the present time developing procedural prompts appears to be
question generation) that precedes the judgments of learning is not an art” (p. 198). Fifteen years later, not much has changed, as
performed until after a delay following the completion of reading very little empirical work has been published to evaluate the
(Thiede, Dunlosky, Griffin, & Wiley, 2005; Thiede et al., 2010). benefits of different procedural prompts (i.e., prompts to stim-
The theoretical interpretation is that delayed study tasks permit ulate generation of different question types). The present ex-
access to more valid, situation-level cues that are then used as a periment thus provides needed empirical evaluation of two
basis for the subsequent judgments (see Thiede, Griffin, Wiley, & previously identified question types (Raphael & Pearson, 1985),
Redford, 2009). Accordingly, the absence of a benefit for relative and importantly indicates that the benefits are nuanced. Within
monitoring accuracy in the present study may reflect the use of a a single experiment, we observed a mixed pattern such that the
delayed judgment procedure in conjunction with a study task benefits of question generation and answering on memory and
performed immediately upon completion of reading. metacomprehension were selective. Following appeals in the
To take stock, our results suggest that the benefits of self- literature (Rosenshine et al.; Wong, 1985), we have proposed a
generating and answering questions on memory for text are at least in contextual approach drawn from basic memory theory (Jenkins,
part determined by the constellation of types of questions generated 1979; McDaniel & Butler, 2011) in an attempt to forge a
(prompted for generation) and answered, the nature of the texts, and coherent understanding of how self-questioning and answering
the information targeted by the criterial test. The present results works—when it is beneficial and when it is not. We do not
emphasize the idea that simply focusing students’ attention on content intend to imply that the present experimental findings confirm
by prompting generation and answering of questions is not sufficient this approach but rather that they hint at the fruitfulness of
to produce benefits over rereading alone. Paralleling basic memory simultaneously considering transfer and material appropriate
research (e.g., Einstein et al., 1990; McDaniel et al., 1986; McDaniel, processing dynamics on the benefits of question self-generation
Einstein, & Lollis, 1988), self-generation and answering of questions and answering for improving memory and metacomprehension
appear to be beneficial primarily when attention is focused on pro- of text.
cessing information that would not ordinarily be encoded when read-
ing particular texts (materials), and when a transfer-appropriate crite- References
rial test is administered.
Note that our interpretation has interesting but as yet untested Connor, L. T., Dunlosky, J., & Hertzog, C. (1997). Age-related differences
in absolute but not relative metamemory accuracy. Psychology and
implications with regard to extending the present findings. First,
Aging, 12, 50 –71. doi:10.1037/0882-7974.12.1.50
we are not arguing that generation and answering of conceptual Davey, B., & McBride, S. (1986). Effects of question-generation training
questions is always better. For instance, for texts that do not on reading comprehension. Journal of Educational Psychology, 78,
ordinarily stimulate processing of details (e.g., a folktale), gener- 256 –262. doi:10.1037/0022-0663.78.4.256
ating conceptual questions may not yield significant benefits. Dermody, M. (1988, February). Metacognitive strategies for development
Instead, generation of detailed questions may serve learners better of reading comprehension for younger children. Paper presented at the
930 BUGG AND MCDANIEL

annual meeting of the American Association of Colleges for Teacher McDaniel, M. A., & Einstein, G. O. (2005). Material appropriate difficulty:
Education, New Orleans, LA. A framework for determining when difficulty is desirable for improving
Dunlosky, J., & Lipko, A. R. (2007). Metacomprehension: A brief history learning. In A. F. Healy (Ed.), Experimental cognitive psychology and its
and how to improve its accuracy. Current Directions in Psychological applications (pp. 73– 85). Washington, DC: American Psychological
Science, 16, 228 –232. doi:10.1111/j.1467-8721.2007.00509.x Association. doi:10.1037/10895-006
Dunlosky, J., & Metcalfe, J. (2009). Metacognition. Thousand Oaks, CA: McDaniel, M. A., Einstein, G. O., Dunay, P. K., & Cobb, R. (1986).
Sage Publications. Encoding difficulty and memory: Toward a unifying theory. Journal of
Einstein, G. O., McDaniel, M. A., Owen, P. D., & Cote, N. C. (1990). Memory and Language, 25, 645– 656. doi:10.1016/0749-
Encoding and recall of texts: The importance of material appropriate 596X(86)90041-0
processing. Journal of Memory and Language, 29, 566 –581. doi: McDaniel, M. A., Einstein, G. O., & Lollis, T. (1988). Qualitative and
10.1016/0749-596X(90)90052-2 quantitative considerations in encoding difficulty effects. Memory &
Fletcher, C. R., & Bloom, C. P. (1988). Causal reasoning in the compre- Cognition, 16, 8 –14. doi:10.3758/BF03197740
hension of simple narrative texts. Journal of Memory and Language, McDaniel, M. A., Friedman, A., & Bourne, L. E. (1978). Remembering the
244, 235–244. levels of information in words. Memory & Cognition, 6, 156 –164.
Foos, P. W., Mora, J. J., & Tkacz, S. (1994). Student study techniques and doi:10.3758/BF03197441
the generation effect. Journal of Educational Psychology, 86, 567–576. McDaniel, M. A., Hines, R. J., Waddill, P. J., & Einstein, G. O. (1994).
doi:10.1037/0022-0663.86.4.567 What makes folk tales unique: Content familiarity, causal structure,
Frase, L. T., & Schwartz, B. J. (1975). Effect of question production and scripts, or superstructures? Journal of Experimental Psychology: Learn-
answering on prose recall. Journal of Educational Psychology, 67, ing, Memory, and Cognition, 20, 169 –184. doi:10.1037/0278-
628 – 635. doi:10.1037/0022-0663.67.5.628 7393.20.1.169
Glenberg, A. M., & Epstein, W. (1985). Calibration of comprehension. McNamara, D. S. (2004). SERT: Self-explanation reading training. Dis-
Journal of Experimental Psychology: Learning, Memory, and Cogni- course Processes, 38, 1–30. doi:10.1207/s15326950dp3801_1
tion, 11, 702–718. doi:10.1037/0278-7393.11.1-4.702 McNamara, D. S., Kintsch, E., Butler-Songer, N., & Kintsch, W. (1996).
Griffin, T. D., Jee, B. D., & Wiley, J. (2009). The effects of domain Are good texts always better? Interactions of text coherence, background
knowledge on metacomprehension accuracy. Memory & Cognition, 37, knowledge, and levels of understanding in learning from text. Cognition
1001–1013. doi:10.3758/MC.37.7.1001 and Instruction, 14, 1– 43. doi:10.1207/s1532690xci1401_1
Griffin, T. D., Wiley, J., & Thiede, K. W. (2008). Individual differences, Morris, C. D., Bransford, J. D., & Franks, J. J. (1977). Levels of processing
rereading, and self-explanation: Concurrent processing and cue validity versus transfer appropriate processing. Journal of Verbal Learning and
as constraints on metacomprehension accuracy. Memory & Cognition, Verbal Behavior, 16, 519 –533. doi:10.1016/S0022-5371(77)80016-9
36, 93–103. doi:10.3758/MC.36.1.93 Noordman, L. G. M., Vonk, W., & Kempff, H. J. (1992). Causal inferences
Jenkins, J. J. (1979). Four points to remember: A tetrahedral model of during the reading of expository texts. Journal of Memory and Lan-
memory experiments. In L. S. Cermak & F. I. M. Craik (Eds.), Levels of guage, 31, 573–590. doi:10.1016/0749-596X(92)90029-W
processing in human memory (pp. 429 – 446). Hillsdale, NJ: Erlbaum. Raphael, T. E., & Pearson, P. D. (1985). Increasing students’ awareness of
King, A. (1989). Effects of self-questioning training on college students’ sources of information for answering questions. American Educational
comprehension of lectures. Contemporary Educational Psychology, 14, Research Journal, 22, 217–235.
1–16. Roediger, H. L., III, Weldon, M. S., & Challis, B. H. (1989). Explaining
King, A. (1992). Facilitating elaborative learning through guided student- dissociations between implicit and explicit measures of retention: A
generated questioning. Educational Psychologist, 27, 111–126. doi: processing account. In H. L. Roediger III & F. I. M. Craik (Eds.),
10.1207/s15326985ep2701_8 Varieties of memory and consciousness: Essays in honor of Endel
Labercane, G., & Battle, J. (1987). Cognitive processing strategies, self- Tulving (pp. 3– 41). Hillsdale, NJ: Erlbaum.
esteem, and reading comprehension of learning disabled students. B. C. Rosenshine, B., Meister, C., & Chapman, S. (1996). Teaching students to
Journal of Special Education, 11, 167–185. generate questions: A review of the intervention studies. Review of
Levy, B. A. (1981). Interactive processes during reading. In. A. M. Lesgold Educational Research, 66, 181–221.
& C. A. Perfetti (Eds.), Interactive processes in reading (pp. 1–35). Smith, N. J. (1977). The effects of training teachers to teach students at
Hillsdale, NJ: Erlbaum. different reading ability levels to formulate three types of questions on
Maki, R. H. (1998). Predicting performance on text: Delayed versus reading comprehension and question generation ability (Unpublished
immediate predictions and tests. Memory & Cognition, 26, 959 –964. doctoral dissertation). University of Georgia.
Maki, R. H., & Berry, S. L. (1984). Metacomprehension of text material. Son, L. K., & Metcalfe, J. (2000). Metacognitive and control strategies in
Journal of Experimental Psychology: Learning, Memory, and Cogni- study-time allocation. Journal of Experimental Psychology: Learning,
tion, 10, 663– 679. doi:10.1037/0278-7393.10.4.663 Memory, and Cognition, 26, 204 –221. doi:10.1037/0278-7393.26.1.204
Maki, R. H., Foley, J. M., Kajer, W. K., Thompson, R. C., & Willert, M. G. Thiede, K. W., Anderson, M. C. M., & Therriault, D. (2003). Accuracy of
(1990). Increased processing enhances calibration of comprehension. metacognitive monitoring affects learning of texts. Journal of Educa-
Journal of Experimental Psychology: Learning, Memory, and Cogni- tional Psychology, 95, 66 –73. doi:10.1037/0022-0663.95.1.66
tion, 16, 609 – 616. doi:10.1037/0278-7393.16.4.609 Thiede, K. W., Dunlosky, J., Griffin, T. D., & Wiley, J. (2005). Under-
Mayer, R. E. (2003). Learning and instruction. Upper Saddle River, NJ: standing the delayed-keyword effect on metacomprehension accuracy.
Merrill Prentice Hall. Journal of Experimental Psychology: Learning, Memory, and Cogni-
McDaniel, M. A., & Butler, A. C. (2011). A contextual framework for tion, 31, 1267–1280. doi:10.1037/0278-7393.31.6.1267
understanding when difficulties are desirable. In A. Benjamin (Ed.), Thiede, K. W., Griffin, T. D., Wiley, J., & Anderson, M. C. M. (2010).
Successful remembering and successful Forgetting: A festschrift in Poor metacomprehension accuracy as a result of inappropriate cue use.
honor of Robert A. Bjork (pp. 175–199). New York, NY: Taylor & Discourse Processes, 47, 331–362. doi:10.1080/01638530902959927
Francis. Thiede, K. W., Griffin, T. D., Wiley, J., & Redford, J. (2009). Metacog-
McDaniel, M. A., & Einstein, G. O. (1989). Material appropriate process- nitive monitoring during and after reading. In D. J. Hacker, J. Dunlosky,
ing: A contextualist approach to reading and studying strategies. Edu- & A. C. Graesser (Eds.), Handbook of metacognition in education (pp.
cational Psychology Review, 1, 113–145. doi:10.1007/BF01326639 85–106). New York, NY: Routledge.
 GENERATION AND ANSWERING OF QUESTIONS 931

Thomas, A. K., & McDaniel, M. A. (2007). The negative cascade of Wong, B. Y. L. (1985). Self-questioning instructional research: A review.
incongruent generative study-test processing in memory and metacom- Review of Educational Research, 55, 227–268.
prehension. Memory & Cognition, 35, 668 – 678. doi:10.3758/
BF03193305
Weinstein, Y., McDermott, K. B., & Roediger, H. L. (2010). A comparison
of study strategies for passages: Re-reading, answering questions, and Received March 6, 2011
generating questions. Journal of Experimental Psychology: Applied, 16, Revision received March 27, 2012
308 –316. doi:10.1037/a0020992 Accepted April 10, 2012 䡲

New Editors Appointed, 2014 –2019

The Publications and Communications Board of the American Psychological Association an-
nounces the appointment of 4 new editors for 6-year terms beginning in 2014. As of January 1,
2013, manuscripts should be directed as follows:
● Journal of Experimental Psychology: Animal Behavior Processes (http://www.apa.org/pubs/
journals/xan/), Ralph R. Miller, PhD, Department of Psychology, SUNY-Binghamton
● Journal of Experimental Psychology: Applied (http://www.apa.org/pubs/journals/xap/), Neil
Brewer, PhD, School of Psychology, Flinders University
● Neuropsychology (http://www.apa.org/pubs/journals/neu/), Gregory G. Brown, PhD, ABPP,
UCSD School of Medicine and Veterans Affairs San Diego Healthcare System
● Psychological Methods (http://www.apa.org/pubs/journals/met/), Lisa L. Harlow, PhD, De-
partment of Psychology, University of Rhode Island
Electronic manuscript submission: As of January 1, 2013, manuscripts should be submitted
electronically to the new editors via the journal’s Manuscript Submission Portal (see the website
listed above with each journal title).
Current editors Anthony Dickinson, PhD, Wendy A. Rogers, PhD, Stephen M. Rao, PhD, and Mark
Appelbaum, PhD, will receive and consider new manuscripts through December 31, 2012.