🏠
Author: mzloteanu.bsky.social (did:plc:oxbw4yyfdujd2rhakdsrg73m)

Record🤔

uri:
"at://did:plc:oxbw4yyfdujd2rhakdsrg73m/app.bsky.feed.post/3l2dbfhos6x23"
cid:
"bafyreicdo67qlzvs2ecyuulsbmp7swwstv5n3ruey4agtqwwymwezzbmwi"
value:
text:
"#statstab #164 Ordinal regression models to analyze Likert scale data

Thoughts: One of the clearest tutorial for ordinal, cumulative (probit) models I've seen. Reports probabilities and expected mean ratinga w/ plots!

#ordinal #brms #likert #probit #r

dibsmethodsmeetings.github.io/ordinal-regr..."
$type:
"app.bsky.feed.post"
embed:
$type:
"app.bsky.embed.external"
external:
thumb:
View blob content
$type:
"blob"
ref:
$link:
"bafkreihrosdulcjbw3ktbkaz5h3zosmlvuyyvbednl556nc33wud7mhuci"
mimeType:
"image/jpeg"
size:
175294
title:
"Ordinal regression models to analyze Likert scale data"
description:
"Today I am going to present on an alternative way to analyze Likert scale data by using ordinal regression instead of linear regression. But first, why is it even a problem to use linear regression wh..."
langs:
  • "en"
facets:
  • index:
    byteEnd:
    9
    byteStart:
    0
    features:
    • tag:
      "statstab"
      $type:
      "app.bsky.richtext.facet#tag"
  • index:
    byteEnd:
    228
    byteStart:
    220
    features:
    • tag:
      "ordinal"
      $type:
      "app.bsky.richtext.facet#tag"
  • index:
    byteEnd:
    234
    byteStart:
    229
    features:
    • tag:
      "brms"
      $type:
      "app.bsky.richtext.facet#tag"
  • index:
    byteEnd:
    242
    byteStart:
    235
    features:
    • tag:
      "likert"
      $type:
      "app.bsky.richtext.facet#tag"
  • index:
    byteEnd:
    250
    byteStart:
    243
    features:
    • tag:
      "probit"
      $type:
      "app.bsky.richtext.facet#tag"
  • index:
    byteEnd:
    253
    byteStart:
    251
    features:
    • tag:
      "r"
      $type:
      "app.bsky.richtext.facet#tag"
  • index:
    byteEnd:
    300
    byteStart:
    255
    features:
createdAt:
"2024-08-22T18:29:29.218Z"