A 10-day experience sampling dataset on subjective experiences of middle and upper secondary school students in 2022

Sample

The dataset was generated from a sample of middle and upper secondary school students in Finland in fall 2022. The dataset comprises a start-up survey as well as momentary self-reports from students over a period of 10 weekdays. Initially, 358 participants completed the start-up survey. Of these, 324 participants participated in the 10-day ESM period. Participants who did not provide consent or did not complete at least three ESM assessments, along with students who had not yet turned 15, were excluded from the dataset, leading to an analytic sample of 302 participants. To note, the raw data file contains approximately 8,260 rows for these 302 participants due to the data’s long format, where each row represents an individual measurement point. For the purpose of participant anonymity, background variables such as age, socio-economic status, school or municipality are not included at the individual level in the dataset available for open access. For the same reason, gender response options “other” and “I don’t want to reply” were collapsed into missing data.

Procedure

Data was collected using the “RealLife Exp” app, developed by LifeData (accessible at https://www.lifedatacorp.com/). During start-up sessions at schools, participants were in large groups with teachers present given instructions on how to download the app “RealLife Exp” on their own mobile phones, including instructions on how to start the survey. The participants were given the possibility to ask any questions related to the data collection and the research project during the start-up sessions. While LifeData operates as a Software as a Service (SaaS) platform and does not issue traditional copyright licenses, our use of the app is fully compliant with their terms of service.

Measured variables

The public dataset described in this data note comprises both baseline and momentary data (Söderberg & Mölsä, 2024).

Start-up survey

The start-up survey includes measures on (a) academic self-efficacy (ASE, a subscale within the Self-Efficacy Questionnaire for Children, SEQ-C; Muris, 2001), (b) school-related social support from peers (Peer Support at School; PSS), teachers (Teachers-Student Relationship; TSR), and family (Family Support for Learning; FSL) (from SEI; Appleton et al., 2006), and (c) general self-efficacy (from S-GSE; Schwarzer & Jerusalem, 1995). To note, a modified version of the S-GSE with a Swedish translation by Koskinen-Hagman et al. (1999) was used. In addition, there is a single item measuring school enjoyment (self-designed), and another item assessing school absenteeism (self-designed).

Regarding academic self-efficacy, four out of the original eight items in the ASE subscale (Muris, 2001) were selected for inclusion in this dataset. Similarly, not all the original 35 items in the SEI (Appleton et al., 2006) were used; instead, the measurement instrument was modified to contain a total of 19 items, distributed across the three subscales. Moreover, from the original S-GSE scale (Schwarzer & Jerusalem, 1995) of ten items, five items of the S-GSE were selected to be included in this dataset. Additionally, for the S-SGE scale, the response scale was modified from a 4-point to a 5-point scale. Apart from the item on school absenteeism, all baseline items were chosen to be rated on a slider scale with two end points, ranging from 1 (“No, not at all”) to 5 (“Yes, absolutely”). The survey item measuring school absenteeism (“Have you been absent from school in the past two weeks?”) consists of three response options: (a) No, (b) Yes, once, and (c) Yes, several times.

Experience Sampling Measures (ESM)

The ESM part was designed to include three blocks of inquiries: a morning block, a pre-and afternoon block, and an evening block. The morning block (Block A) encompassed three items: (i) “Good morning. Did you sleep well last night?”, (ii) “Have you eaten (or will you eat) breakfast this morning?”, and (iii) “Do you feel okay about going to school today?”. The pre- and afternoon block (Block B) aimed to assess the respondent’s current emotional state with the query (i) “How are you feeling right now?” (8 items), (ii) perceptions of the most recent lecture with the query “The recent lecture was …” (6 items), and (iii) an assessment of their enjoyment during peer interactions with the query “Did you have a good time with your classmates in the morning/in the afternoon?” (1 item). Lastly, the evening block (Block C) was designed to assess overall school day satisfaction with a single query: “What rating would you give for the overall school day?”.

The first query within Block B includes items based on both positive and negative emotions following the PANAS framework (Watson & Clark, 1994; Watson et al., 1988). All other ESM items were designed by the project. All ESM items were selected to be rated on a 5-point slider scale (1: No, not at all, 5: Yes, absolutely), except for the item within Block C, which was rated on a 10-point slider scale.

Translation of items

In addition to utilizing existing modified versions of scales with Swedish translations, such as the S-SGE scale, all items within the dataset were translated into Swedish for the data collection. The translations were refined within the research group and confirmed by the project leader. Prior to data collection, the translated items underwent testing through a series of pilot studies conducted among secondary school students, with participant numbers ranging from three to fifteen for each test.

ESM protocol

The ESM protocol encompasses four measurement points per day during a period of 10 school days across two weeks (Monday-Friday), resulting in a total of 40 measurement points. The ESM surveys were administered at three designated fixed intervals (07:15 a.m., 09:45 a.m., 01:05 p.m.) and one randomized interval (07:00–10:00 p.m.), with a 1.5-hour response window provided for each prompt. Respondents had the option to skip any question but could not revisit or modify their response.

To mitigate potential reporting fatigue, the assessment scales were kept brief. Additionally, a planned missingness approach was implemented for the PANAS items to mitigate systematic non-response effects. Specifically, one item (in block B: “I enjoy being at school”) was consistently included, while each of the other PANAS items was intentionally omitted on one out of seven prompts. Furthermore, additional signals with random questions unrelated to the dataset, were prompted once every afternoon to add variety. Respondents were regularly notified of their progress throughout the data collection period by means of in-app information. As external motivation, respondents who participated in both the baseline survey and the ESM period were rewarded with a 20 € gift card. See Table 1 for an overview of the ESM protocol designed for this dataset.

Dataset validation

Pilot studies

To ensure face validity and to assess participant burden, a series of four pilot studies were conducted among secondary school students prior to the main data collection. The sample sizes for each pilot ranged from three to 15. As part of these pilot studies, participants were prompted at the end of the ESM session to answer a few questions about their experiences with the study. Additionally, a subset of the participants agreed to participate in brief (10-minute) online video interviews to provide more detailed feedback on their experiences. The overall feedback from the pilots indicated that participants thought that participation in this type of study was meaningful, especially when compared to other types of studies such as cross-sectional surveys commonly conducted within schools. The pilot studies also confirmed that participants understood the items as intended, thus affirming the face validity of the measures.

Factor analysis

To assess construct validity, a confirmatory factor analysis (CFA) was performed using the SEQ-C, SEI, and S-GSE scales from the baseline survey. Prior to the CFAs, the KMO values indicated adequate sampling adequacy for all scales: SEQ-C (KMO = 0.736), SEI (KMO = 0.902), and S-GSE (KMO = 0.843), as recommended by Kaiser (1974). Additionally, Bartlett’s Test of Sphericity (BTS) was significant for each scale (p < 0.001), indicating that the variables were significantly correlated and justifying the use of the factor analytic model.

Maximum Likelihood (ML) estimation was employed for the CFA of all scales, except for the SEI scale, where robust Maximum Likelihood (MLR) was used due to high kurtosis (>3) in one of the items. The individual CFAs indicated an overall adequate model fit, though the RMSEA for the SEQ-C scale fell outside the acceptable range. For detailed goodness-of-fit statistics, see Table 2. The analyses of KMO and BTS were conducted using IBM SPSS Statistics 29, while the CFAs were performed using MPlus Version 8.8.

Dataset reliability

To evaluate how consistently the items within each scale measured the same underlying construct, i.e. between-person reliability, internal consistency checks for this dataset were performed using Cronbach’s alpha (Cronbach, 1951). Sufficient internal consistency was displayed for the three sub-scales within the SEI, with α = 0.88 (PSS), α = 0.92 (TSR), and α = 0.89 (FSL), for the S-GSE scale (α = 0.84), and for the SEQ-C scale (α = 0.76).

Limitations

Firstly, as the dataset relies on self-report measures, there is always a potential for response bias or social desirability effects. However, measures were taken to minimize these biases through anonymous data collection, clear instructions provided to respondents as well as a planned missingness design. Secondly, while the sample size of the dataset can be considered sufficient for many analyses, a larger sample size would allow for more robust and generalizable results.