Method

A qualitative, repeated-session observational design was used to investigate participants’ cognitive processes during basketball performance. Data were collected using a concurrent Think Aloud methodology, in which participants verbalised their thoughts while completing attacking and defensive basketball tasks (Eccles & Arsal, 2017). This methodology was selected because it allowed thoughts relating to attentional focus and other psychological constructs to be recorded as they occurred during performance, rather than relying solely on participants’ retrospective recall. 

The procedure most closely resembled a Level 2 Think Aloud protocol because participants verbalised immediate thoughts, perceptions and reactions without being consistently required to explain or justify them. Think Aloud has increasingly been used within sport and exercise psychology to investigate cognitive processes during performance, although researchers must clearly describe how verbalisation was prompted, recorded and analysed (McGreary et al., 2024). Its use in applied sport settings is also demonstrated by Quick and Lyle (2024), who used an in situ Think Aloud protocol to investigate cognitions that occurred during sporting practice. 

Participants were four undergraduate students; all had previous basketball experience. Two participants were classified as amateur players with limited informal experience, while the other two were classified as more experienced players because they had previously competed in club basketball. Participants were recruited through convenience sampling because the investigation was conducted as part of scheduled university practical classes. 

Before data collection, participants were informed about the basketball activity, the requirement to verbalise their thoughts during the performance, and the use of the video recordings for the group assessment. Participants were also informed that they would assist in transcribing, coding, and analysing their own recorded verbalisations. 

After completing a pilot session and familiarising participants with the tasks, we collected data from over four weekly basketball games during the data collection period. During each session, each participant completed two different types of data collection: offensive and defensive. A total of 32 data recordings were planned for the four basketball participants; however, due to some participants being absent and the loss of recordings caused by camera malfunctions (e.g., recordings not being saved), only 28 recordings, both offensive and defensive, were included in the final data set used for analysis. 

These tasks involved playing a half-court basketball game with four offensive players vs. three defensive players. Each team spent seven minutes attacking the opponent and seven minutes defending their basket. In the offensive situation, the goal for each participant was to create scoring opportunities using passing, dribbling, movement, positioning, and shooting. In contrast, in the defensive situation, each participant attempted to create an opportunity for their opponent to score by applying pressure, marking an offensive player, blocking shots and clearing the area for a rebound.  

When a basket was scored or play was stopped, the attacking team returned the ball to approximately halfway before restarting. Players in defence who were actively involved would rotate into play after resets. All participants with a camera were to verbalise their thoughts immediately before or while executing their actions. Some of these thoughts included calls for possession, identification of passing options, communication of court position, directions to apply defensive pressure and responses to shooting or rebounding opportunities. 

Although the activity's general purpose remained the same, the conditions were not fully standardised across the four sessions. Factors such as rotation procedures, game duration, restart procedures and enforcement of the verbalisation rule were not the same for every week. These differences were also considered when analysing patterns across the different sessions.  

First-person video and audio data were collected using head- and chest-mounted GoPro cameras. The cameras recorded each participant’s view of the court, while the microphone captured their concurrent verbalisations. Combining the audio with first-person video allowed each statement to be interpreted in relation to the basketball action taking place at that moment. 

At the end of each session, the recordings were transferred to local photo and video folders on the group members’ laptops. Files were labelled according to the participant. week and playing condition, using titles such as “Week 3 Offence” and “Week 3 Defence.” Access to the footage was limited to members of the assessment group. 

The recordings were analysed using semantic coding and thematic analysis, guided by the general process described by Braun and Clarke (2006). Semantic coding was used because the initial analysis focused on the explicit meaning of each spoken statement rather than attempting to infer an unspoken or underlying meaning. 

After each session, the recordings were viewed several times. Audible verbalisation was transcribed verbatim, including short statements, repeated words, pauses and informal expressions. The transcripts were organised into a table with four headings: timestamp, verbalisation, context, and code. 

The timestamp identified when the verbalisation occurred in the recording. Verbalisation contained the participant’s exact words, while context briefly described the basketball event that was occurring at that moment. The code provided a concise description of the meaning or function of the statement. 

For example, statements such as "pressure", "block", and “rebound” were initially coded according to the immediate defensive or attentional action they represented. Statements such as “here” and “middle” were coded as calling for possession and for communicating position or court awareness. Statements such as “bad shot", “my bad", and “good pass” were coded as evaluations of successful or unsuccessful performance. 

After initial coding, codes with similar meanings were compared and grouped into broader categories and themes. This was an iterative process in which the researchers moved between the original videos, transcripts, initial codes and developing thematic structure. Themes were not developed only according to how frequently a statement occurred; the relevance and meaning of each pattern were also considered. 

The initial attention-related data included verbalisation concerning task-relevant game cues, player and court positioning, offensive and defensive communication, time awareness, physical fatigue, outcome monitoring and attention directed towards previous errors. Final themes were retained when their sub-themes and supporting statements formed a coherent pattern that was meaningfully different from the other themes.  

The study used numerous procedures to improve its accuracy and reliability. For example, all recordings were reviewed several times to allow all transcripts to be verified against both the audio and the visual context. This process also helped prevent the misinterpretation of short verbalisation by considering the activity that had occurred simultaneously. 

Secondly, verbalisation time stamps were logged so that statements, codes and themes could each be traced back to their respective original video files. All participants used the same transcription/coding table throughout all weeks and playing conditions. The result was an audit trail that created a connection between the original video data and the final thematic results. 

In addition, several other team members verified and agreed on the appropriateness of the codes. After agreeing on the previously identified codes and the most appropriate broad theme, shared definitions were developed for each frequently used code. For example, the team developed shared definitions for positive team encouragement, uncertainty in decision-making, physical preparation, body awareness, and positive expectations. In doing so, we increased the consistency of coding within our team and increased the likelihood that the identified themes did not represent just one researcher's perspective.