srearl/cslqualtrics

R-based tools to format Qualtrics observation data

cslqualtrics's Issues

address formatting inconsistencies

Unfortunately, the new program delivers an output data file that is not identical to the old format and as such it will not add the downloaded data to the large existing Query Data file.

For example: the entire ‘Other’ column has NA listed in it and NA is used for missing data cells… they should just be left blank. Also – there are numbers entered in cells instead of names for the behavior categories…

And, I think this matters as well, but the ‘Date’ column is not in the same place…

The two files will not merge. I think it’s because the cells don’t contain exactly the same formatted data.

I am attaching a copy of the final cleaned output file using your new instructions. And the latest cleaned data file so you can see the differences (Go to Merge view). You can also compare the new generated cleaned data file to the Query data file you have on your thumb drive, you will see the differences.

why are summer data from only 2015 and 2016 catalogued in the database?

additional data resources requested

In addition to the behaviour graph, Anne would like also ready access to:

• child-child interaction data: which children are interacting, and in what way?
• TA-child interaction data: are students responding differently to TAs, quantify behaviour(s) relative to TA
• behaviour by location graph
• TA stats: how many entries are TAs logging, and, ideally, what type of entries are they logging (e.g., are they logging more bad relative to good behaviours)?

docs: behaviour plot using echarts4r

obs <- DBI::dbGetQuery(
  conn = csl_obs_db,
  statement = "
  SELECT
  response_id,
  question,
  response
  FROM
  observations
  WHERE
  survey_id = 'SV_6W59cgkw9TdB6BM'
  ;
  "
)

jack_ids <- obs[
obs$response == "Jack B." &
grepl("^q2_", obs$question, ignore.case = TRUE),
]$response_id

jack_data <- obs[obs$response_id %in% c(jack_ids), ]

jack_behaviours_sum <- jack_data |>
  dplyr::filter(grepl("^q4$", question, ignore.case = TRUE)) |>
  nrow()

student_behaviours_sum <- jack_behaviours_sum

devtools::load_all("~/localRepos/cslqualtrics/")

jack_data |>
  dplyr::filter(grepl("^q4$", question, ignore.case = TRUE)) |>
  dplyr::group_by(response) |>
  dplyr::summarise(
    n       = dplyr::n(),
    percent = round((n / student_behaviours_sum) * 100)
  ) |>
  dplyr::ungroup() |>
  dplyr::full_join(
    y  = cslqualtrics::behaviour_types,
    by = c("response" = "behaviour")
  ) |>
  dplyr::arrange(
    type,
    response
  ) |>
  echarts4r::e_charts(x = response) |>
  echarts4r::e_bar(
    serie   = percent,
    name    = "student behaviour",
    legend  = FALSE,
    colorBy = data
  ) |>
  echarts4r::e_tooltip(trigger = "axis") |>
  echarts4r::e_x_axis(axisLabel = list(interval = 0, rotate = 315)) |>
  echarts4r::e_title("Jack") |>
  echarts4r::e_grid(bottom = "40%") |>
  echarts4r::e_color(color = c(cslqualtrics::behaviour_types$color)) |>
  echarts4r::e_y_axis(max = 60)

behaviour graph modifications

• fix x axis to show all behaviours
• fix y axis to 60%
• order and color-code behaviours according to behaviour type (e.g., positive, negative); Anne will provide the categories

why are there not any observation data for summer 2023 or fall 2023 reliability?

...because there are not any

historic data in design considerations

Hi Anne,

Trying, though not terribly successfully, to work it in currently.

A couple of questions for you that will help inform the design:

• When the approach that you had been using was working, data sheets from many years and semesters were aggregated such that you could graph data from the current year and semester or those in the past. Is that a feature that you need? My impression is that you are using this tool primarily in your conversations with parents. As such, is there value in being able to graph data from, for example, the spring of 2017?

We use these data for parents, but also for research purposes. For example: we are tracking the development of Theory of Mind in children and how this behavior correlates to both problem behavior and to social skills. We need to follow a specific child over their enrollment at the CSL correlating their observed behavior to assessment outcomes.

• We have focused on the Behaviour_Graph tab but there are other plots (Overall Location, Location by Behaviour), and data summaries (e.g., TA_Text) in the spreadsheet. Are the plots and summaries other than the Behaviour_Graph helpful? Are there any other data that you would like visualized or summarised that are not reflected currently?

I was waiting to be able to use those other data as well!

We routinely need to know how many data entries were entered by each TA – this is part of their performance evaluations each semester. Also – it would be wonderful to know in which of the setting categories the child’s behavior occurs – e.g., are most inappropriate behaviors more likely to occur during transitions or large group time, etc. The setting is very valuable in determining the behaviors antecedent stimulus – i.e. under what condition is the behavior more likely to occur.

Also, who is the child targeting – what other child – both socially (e.g., who does the child share a toy with) and antisocially (e.g., who does the child aggress towards) to see if there are any patterns there.