• Is there a way to delay the onset of the the response time measurement, until after the web application has fully loaded?
• Alternatively... the application records and displays the loading time, in the format "Data loaded in XX settings", when 'Details' is clicked on the loading screen. Might there be a way to extract the number of seconds here, and automatically subtract it from the response time for the question?

holch wrote: I hope you are aware that not everyone might be able to access this application. Like me?

77ms: CE Web Viewer version: 5.1 [ 5.1.0.201702230534 | html:5.1.0.201702230534_ ]
88ms: GL detect: WebGL 1.0 More
4041ms: Loading webscene id : 6eaa9c2dfcab4bb0bb5276692e2f8f03
4054ms: Fallback to custom unzip on arraybuffer and custom streamParser
41597ms: Downloading: 100% (9.14 of 9.14 MB)
41598ms: Requested gzipped data, got gzipped data
41600ms: Using custom unzip and streamparser on arraybuffer
54015ms: Reading: 98% (9 of 9.14 MB)
54260ms: Data loaded in 51 seconds
54378ms: Initializing...
54384ms: Objects: 7, Materials: 7, Textures: 2, Geometries: 7, Layers: 6
54385ms: Web Scene version: 2.0 (Esri ArcScene (Build 4959))

Since security is getting tighter in browser is there a way to have a listener, which scraps the frame for content? Is that possible with moderns browsers and tight security settings around iframe.

<script type="text/javascript" charset="utf-8">  
 
    // Record a timestamp when page loads
    $(document).ready(function(){
      $('#question{QID} input[type="text"]:eq(0)').val(new Date);
 
    });  
 
    // Function to run when message received
    function receiveMessage(event) {
      // Quit if message not from expected origin
      if (event.origin !== 'https://theIframeDomain.com') { 
        return;
      }
      // It's a valid origin...
      else {
        // Verify the content of the received message
        if(event.data.indexOf('Some identifying string in data') < 0) {
          return;
        }
        else {
          // Record a second timestamp
          $(document).ready(function(){
            var messageDate = event.data.replace(/Loaded time:/, '');
            $('#question{QID} input[type="text"]:eq(1)').val(new Date);      
          });
        }
      }
    }
    // Message listener
    window.addEventListener("message", receiveMessage, false);
</script>

jelo wrote:

holch wrote: I hope you are aware that not everyone might be able to access this application. Like me?

At least you didn't had to wait 54 sec

There are some hardware requirements for the graphic card and driver as well.
doc.arcgis.com/en/arcgis-online/referenc...427EA5D97A14BE32737A

So that might explain why Win10 and Chrome were not enough.

tpartner wrote: - have the iframe send a message when the application is completely loaded

tpartner wrote: My approach would be:

- load input-1 of a multiple-text question with a timestamp when the survey page loads
- have the iframe send a message when the application is completely loaded
- upon receipt of the message, input-2 of the multiple-text question is loaded with a new timestamp

<style type="text/css">.arcapp { display:none;}
</style>
<div class="arcapp">
<h2>3D App goes here</h2>
</div>
<script type = "text/javascript">
$(document).ready(function() {
$(".arcapp").delay(60000).fadeIn(500);
});
</script>

sheffieldthesismaps wrote: Do you mean a multiple short-text question?

Tpartner is not talking about your question you want to ask. The javascropt has to save the timestamp somewhere (textfield).

sheffieldthesismaps wrote: Another potential solution I thought of, is to hide all content for 60 seconds with a secondary 'loading screen'', and then just subtract 60 seconds from the recorded response time. Perhaps an easier/more workable solution?

What about users which are not seeing the ArcGIS viewer after 60 seconds? Cause the loading took longer? Environment of every respondent can differ too much to interpret duration in a coherent way.

sheffieldthesismaps wrote: Is there an simple way to delay timing for 60 seconds, or to subtract 60 seconds within lime survey, instead of doing it later/manually in excel/spss, when I analyse the data?

It's ain't getting simpler then doing a SPSS subtraction.