Searching for good mood: Examining relationships between search task and mood



We analyzed mood changes before, during and after online information search and examined effects of search tasks' type on the mood. Searchers' mood was measured using the Positive Affect and Negative Affect Scale (PANAS). Search tasks were characterized by the topic, sequence in a search process, difficulty level, and searcher's interest in a task. Our analysis indicates that participants' mood remained stable during the course of the search and was not affected by the search task topic, sequence in a search process, difficulty level and searchers' interest level. Our findings suggest that searching is a complex experience and factors outside of the search task domain may contribute to searcher's mood.

Introduction and Background

There has been an increased interest in affective dimensions of human computer interaction (Picard, 1997; Julien, McKechnie, & Hart, 2005; Nahl & Bilal, 2007). One of the earlier studies that investigated relationships between searchers' feelings, thoughts and actions during various information search stages was the work of Carol Kuhlthau (Kuhlthau, 1991). Since then, a large number of studies have examined relationships between system-related (e.g., system features and performance) and user-related (e.g., cognition and affect) dimensions of searching. Some of these studies examined relationships between the nature of a search task, its difficulty, searcher's interest in a task, and searcher's affective experiences during or after search. We conducted a study to validate some of the previous findings and examined relationships between the nature of a search task and searcher's mood reported during the search.

Library and Information Science literature have identified a number of factors that influence feelings experienced during online searching. Some of these factors include system design and system performance (Bilal, 2000; Bilal & Bachir, 2007; Kalbach, 2006; Lazar, Jones, Hackley, & Shneiderman, 2006); searcher's performance (Wang, Hawk & Tenopir (2000) and successful task completion (Bilal & Kirby, 2002). Affective states have also been shown to be influenced by the nature of the task and perceptions of task difficulty, user's interest in a search process or a document, and moods and attitudes prior to and during the search.

In a study of the effects of computer apologies, Tzeng (2004) found that the game's difficulty level was the most important predictor of participants' satisfaction with the game. Playing an easier game resulted in a better overall performance and generated feelings of gratification, confidence in playing the game and confidence in future success. Perceived difficulty was found to influence uncertainty, expected effort and motivation to complete the task in the study of senior college students' information behavior (Nahl, 2005). Arapakis, Jose and Gray (2008) examined changes in searchers emotions during performance of various tasks. The authors reported a progressive transition from positive to negative valence as the degree of task difficulty increases. In a study that examined the role of subjective variables in a search process, task difficulty (judged by searchers) was correlated with higher satisfaction levels after the search (Gwizdka & Lopatovska, in press).

Another factor that was shown to influence affect is interest in a search process or a document. Kracker (2002) and Kracker and Wang (2002) studied effects of educating students about Kuhlthau's Information Search Process model and found that positive emotions were associated with the interest in a search process and documents. A study of affective valuation of electronic documents (Lopatovska & Mokros, 2007) found that interest in a document and document's stylistic properties influenced participants' positive and negative feelings.

Finally, affective experiences before and during the search has been linked to the feelings experienced after the search. In a study of frustrating computer interaction, Lazar Jones, Hackley and Shneiderman (2006) found that frustration levels during interaction were negatively correlated with the mood after the session, indicating that negative experiences during the search resulted in a negative mood after the search. Gwizdka and Lopatovska (in press) found that positive feelings expressed before the search task were linked to the positive feelings after the search task. In a study of affective motivation during the online information search, Nahl (2004) found positive correlation between self-efficacy and optimism, and motivation for completing the task. The author found that higher self-efficacy and optimism were associated with higher satisfaction.

The previous research has led us to believe that the nature of search task and its difficulty level contribute to the mood experienced during the search. We also found evidence that the mood prior to the search and interest in the search task influence searchers' feelings. We conducted a study to test the hypothesis that a search task topic, task sequence, task difficulty level, searchers' interest level in the task, and mood prior to the search affect mood reported after the completion of search tasks. Based on Arapakis et al. study (2008), we hypothesized that more difficult tasks would result in an increase of negative mood, while easier tasks will lead to an increased positive mood. Based on some earlier studies, we also hypothesized that interest in a search task would be positively correlated with the positive mood (Lopatovska & Mokros, 2007). The article describes the study design and reports the study findings.




Participants' mood was measured using Positive Affect (PA) and Negative Affect (NA) Schedule (PANAS) (see Appendix A). The PANAS comprised of two 10-item scales that measure positive affect (extent to which a person feels enthusiastic, active, alert, etc.) and negative affect (extent to which a person experiences subjective distress, including anger, contempt, disgust, guilt, fear, nervousness, etc.). The PANAS has demonstrated high reliability and internal and external validity; it is brief and easy to administer (Watson, Clark, & Tellegen, 1988). The instrument is frequently used in psychology and other fields to measure past and present moods (Crawford & Henry, 2004; Mackinnon, Jorm, Christensen, Korten, Jacomb, & Rodgers, 1999; Thompson, 2007). Numeric responses on a scale of 1 to 5 to the ten PA and ten NA items were added to derive individual PA and NA scores ranging between 10 and 50. PANAS was administered before the search, after the completion of the first and the second search tasks. PA and NA questions presented to participants before the search asked about their mood during the past week; PA and NA questions presented to participants after search tasks asked about their feelings during the previous completed search.

Search tasks

The participants were given two search tasks during the course of the experiment. The two search scenarios were informed by the work of R. W. White (2004) who classified a set of tasks based on complexity levels. Search tasks devised by White were used in other studies (White, Ruthven, & Jose, 2005; Bell & Ruthven, 2004). We pre-tested White's tasks in a pilot experiment and selected the two tasks that were most consistently judged as easy and difficult. The original text of the two search tasks from White study was slightly modified for American English.

Low complexity task was defined as a task that provided subjects with more information on what needs to be found. 'Subjects generally found tasks in this category more ‘clear’ and ‘simple’ than those from other categories (White, 2004, p. 185). Below is the text of the low complexity search scenario used in the study:

A friend has recently been applying to various universities and courses but has been complaining that he finds it difficult to get accepted due to the rising numbers of students. You were unsure if his assessment was correct so you have decided to find out how the size of the student enrollment changed over the last 5 years and how it is expected to change in the coming 5 years.

High complexity task was defined as a vaguely formulated task requiring information from multiple sources. 'Subjects found these tasks difficult and classified tasks in this category as least ‘clear’ and ‘simple’ (White, 2004, p. 185). Below is the text of the high complexity task:

Your friend has just finished reading a copy of a national newspaper in which there is an article about Internet music piracy. The article stresses how this is a global problem and affects compact disc sales worldwide. Unaware of the major effects you decide to find out how and why music piracy influences the global music market.

Search tasks were rotated using a Latin square design (AB, BA, AB, etc.), so that half of the participants received the more difficult task first, another half received the easier task first.

Interest in a task and perceived task difficulty

Data on the perceived task difficulty and interest in the search topics were collected in the post-search interview. Participants were asked to identify which of the two search tasks they found 1) more interesting and 2) more difficult.


The study was conducted during the fall of 2008 in a large north-east university. An experiment took place in a laboratory with a computer desk, two monitors, a keyboard, and video cameras. One monitor was positioned directly in front of the user. The monitor was used for searching and only displayed Google search engine in the Windows Internet Explorer version 7. The second computer monitor was used to display the text of the search instructions, search task scenarios and pre- and post- task questionnaires.


Each participant was scheduled for an individual session lasting from 40 to 120 minutes. Upon arrival to the lab, participants read and signed a consent form and received an explanation of the experimental procedure.

Participants' actions during the session were guided by the online questionnaire that solicited demographic information, presented the text of the two search tasks instructions on how to search and provide answers, and measured participant's affective states before and after each search task. Participants were left alone in the lab during the session and were asked to call for the experimenter when they finished all the steps outlined in the questionnaire.

Participants conducted two searches and filled out the questionnaire before and after each search. Upon survey completion, they called for the experimenter who was waiting in the neighboring room. At the end of the session, experimenter debriefed and interviewed participants.

At the end of the experimental session, participants were given course credit for participation in a study.


Thirty six undergraduate students enrolled in a psychology course participated in the study. Six cases were incomplete and had to be discarded resulting in the thirty complete cases (N=30). Table 1 offers descriptive statistics on the participants.

Table 1. Descriptive statistics on a research sample
original image


Mood changes during the search

On average, participants experienced more positive affect than negative affect (Table 2). Before starting the first search task, participants on average reported significantly higher positive affect than after completing search tasks one and two. Positive (PA) and negative affect (NA) did not seem to be influenced by the search task's topic (Enrollment or Music Piracy) or the difficulty level. For example, participants who researched the Enrollment task first followed by the Music Piracy task received an average PA score of 25 for the first task and 24 for the second, and participants who received the Piracy task first followed by the Enrollment task received an average PA score of 23 for the first task and 22 for the second (Table 2). This observation was further tested by running multiple regression analysis described in the next section.

Table 2. Mood scores reported during the search (PA=positive affect; NA=negative affect)
original image

Figure 1 illustrates the changes in participants' positive and negative affect reported before starting the search and after completion of the first and second search tasks.

Figure 1.

Average changes in positive and negative affect during the search

Mood and task-related variables

We used multiple regression analysis to examine effects of task-related variables and mood prior to the search on the mood reported during the search. Independent variables included NA and PA scores reported prior to the search; task type (an identifier based on task sequence and topic); interest in a task; and perceived task difficulty. These independent variables were used to predict PA and NA scores collected after search tasks' completion. Results of the statistical models were not statistically significant, with R2 =.12, F(5, 59) = 1.440, p =.225 for the model predicting post-search PA scores; and R2 =.06, F(5, 59) =.632, p =.676 for the model predicting post-search NA scores. The results of the statistical model indicate that task-related variables explained 12% of the variance of the PA scores, and 6% of the NA scores variance. Interest in a search task was the strongest predictor of PA scores after search tasks, β ‖ =.25, t = 1.800, p =.078. NA prior to the search was the strongest predictor of NA scores after search tasks, with β ‖=.190, t = 1.335, p = 187.

Discussion and Conclusion

We examined several factors that could have influenced the positive and negative affect experienced during the online search. Our statistical analysis indicated lack of statistically significant relationships between mood prior to the search, topic of the search, sequence of task, task difficulty level, searcher's interest in a task and positive and negative affects. Effect sizes (R<sup>2</sup>) of the tested models were also relatively small. These findings were also confirmed by the post-search interviews with participants. When asked to recall the most emotional moments experienced during their searches, most of the participants could not recall anything. For example, when information was hard to find, participants did not report high frustration levels; when participants found results, they did not indicate an extreme happiness, etc. We also asked participants if performing the search in a laboratory setting impacted the way they searched and behaved. For most of the participants, the lab setting was not at all problematic and did not differ from the way they usually use computers in a library or another public place. So, why did task properties and mood reported prior to the search had such a minimal effect on moods experienced during online searching?

One possible explanation is well supported by the human information behavior literature: searching is a complex experience involving many variables of potential consequence to the mood (Nahl & Bilal, 2007). We only investigated some of them. Other search-related factors, such as participants' performance, searching skills, motivation to obtain the best possible results can contribute to affect experienced during the search. There might also be factors outside the scope of an online search experiencing influencing search moods. For example, most of our participants reported feeling tired and worried about their mid-term exams. It is possible that while students participated in our experiment and searched the web for the two given tasks, they were thinking and worrying about their exams and other personal issues.

Another possible explanation for the relatively stable mood throughout the search lies in the nature of the mood construct, which is a relatively long lasting feeling that, unlike emotion, is not felt ‘about’ anything (Morris, 1999) and is not easily influenced by the search stimuli.

The nature of the study findings can also be attributed to the study and instruments' design. One can argue that participants did not have a personal stake in the quality of the search outcomes and did not experience extreme positive and negative reactions to the search stimuli. From the beginning of their search, participants knew that they would be awarded research credits for participation, regardless of the outcome quality. It is possible that in a naturalistic setting where participants had to find information to satisfy their personal information need, their behavior and affective profile would be different. The topics of the two tasks might have contributed to the results as well. It would be interesting to observe if the relationships between search task variables and mood change when people are searching for information that has important consequences in their lives of the lives of their loved ones (e.g., health issue). Our results can also be attributed to the measurement instrument error (e.g., participants' fatigue while filling out PANAS questionnaire). Further inquiries into the relationships between the mood and information search can help to address some of these issues.

While the study did not find statistically significant correlation between the mood reported prior to the search, search task characteristics and the mood reported after the completion of search tasks, the findings suggest that online searching is a complex and rich experience and is part of a larger life context. The study used an affect measurement instrument, PANAS that was not previously applied in the library and information science research, and can be used in the future. The future studies of subjective experiences during an information search can benefit from the study findings and methods.