Behavioral Metrics · 10 min read
Unanswered Questions as Text Message Evidence
A question sent and never answered is a measurable event. What that supports in co-parenting and workplace records, and the ways an export makes it look like one when it is not.
The short answer
A question in a message record is detectable, and whether it received a response before the conversation moved on is checkable. Textimony measures this with an analyzer named question_ratio, version 1.3, reporting how many messages from each participant were questions and how many went unanswered within the thread. This is more useful than most behavioral measurements for one specific reason: it maps onto something people actually argue about. In co-parenting disputes the recurring claim is not that someone was hostile but that someone would not answer, about pickup times, medical appointments, school events. That claim is testable against the record in a way that "was uncooperative" is not. The measurement identifies the questions and their fate; a reviewer then reads them and decides whether the pattern is what it appears to be, because an unanswered question in a file is not automatically an ignored question in life.
Detecting a question is harder than it sounds
Punctuation is unreliable in messaging. "what time are you picking her up" is a question with no question mark, and "really?" frequently is not one. A detector keyed only on the character will miss most of the first category and over-collect the second, which is why question detection combines interrogative structure with the mark rather than relying on either. Rhetorical questions are the harder case. "why do you always do this" is grammatically interrogative and pragmatically an accusation, and counting it as an unanswered request for information inflates the measurement in exactly the threads where people are most likely to argue about it. Candidates are reported for reading rather than totalled into a score, because this distinction is not reliably machine-decidable.
What makes an answer count as an answer
A reply is not automatically a response. Textimony treats a question as answered when a subsequent message from the other participant addresses it before the thread moves to another topic, which means a reply that changes the subject leaves the question open. That is deliberate: in disputes about information withholding, the pattern at issue is usually not silence but consistent non-answers. Timing needs a boundary too. A question answered eleven days later, after three intervening topics, is not usefully the same event as one answered in four minutes. Both are recorded, and the interval is reported alongside, so a reviewer can see the difference rather than inherit a binary that hides it.
The export artifacts that fake this pattern
Cross-platform replies are the most common. A question sent by SMS and answered in a parenting app appears unanswered in an SMS export, and a record assembled from one platform will show a run of ignored questions that were answered somewhere the file cannot see. Where both platforms are available, both belong in the case. Phone calls do the same thing more invisibly, because they leave nothing in the message record at all. A thread reading "call me" followed by silence is entirely consistent with a call that answered everything, and nothing in the export distinguishes that from a question that was never addressed. Where call logs exist and are relevant, they belong alongside the messages for exactly this reason. Date-range trimming produces a third artifact: questions near the end of a trimmed export have no opportunity to be answered inside the file. Any unanswered-question count needs the window stated, and questions in the final days of that window should be excluded or flagged rather than counted.
Using it in a parenting or workplace record
The durable version of this analysis is narrow and specific. Rather than an aggregate rate, isolate the questions that concerned a decision with a deadline -- an exchange time, a medical consent, a school form -- and follow each one to its outcome. A small set of concrete, dated, unanswered logistics questions is more informative than a percentage over an entire thread, and considerably harder to argue with. It also holds up better as a summary. Under Rule 1006, a chart drawn from voluminous records rests on the underlying material staying available. Ten questions, each with its date, its text, and what followed, can be checked message by message. An aggregate response rate invites a dispute about the denominator that the number itself cannot settle.
Can unanswered text messages be used as evidence?
The messages and their timestamps are ordinary record content in an authenticated export, and what was asked and what followed can be summarised. Whether a pattern of non-response is relevant is for counsel and the court. The record establishes that a question was sent and what came after it, not why.
What if the question was answered by phone?
Then the message record is misleading on that point, and this is the most important limitation of the measurement. Calls leave nothing in a message export, so a question followed by silence in the file is fully consistent with a call that answered it. Any unanswered-question analysis should say so plainly.
How does Textimony decide a question went unanswered?
The question_ratio analyzer, version 1.3, treats a question as answered when a later message from the other participant addresses it before the thread moves to a new topic. A reply that changes the subject leaves it open, and the interval to any answer is reported alongside.
Are rhetorical questions counted?
They are detected, which is why results are candidates rather than a score. "Why do you always do this" is interrogative in form and an accusation in function, and no reliable automated rule separates the two. A reviewer reads the candidates and decides which were genuine requests for information.
Is a low response rate proof of information withholding?
No. It is a measurement over one platform across one window, subject to calls, cross-platform replies, and trimming at the edges of the export. It identifies a pattern worth examining. What that pattern means requires reading the specific questions and their circumstances.
What is the most defensible way to present this?
A short, dated list of specific questions that concerned a decision with a deadline, each followed to its outcome, with the export and window stated. That form can be verified message by message. An aggregate percentage invites an argument about what was in the denominator.
Published by
Textimony. Editorial status: Source-linked informational guide. Updated: 2026-07-30.
Sources
Federal Rule of Evidence 1006: Summaries to Prove Content; Federal Rule of Evidence 901: Authenticating or Identifying Evidence; California Courts Self-Help: Custody and parenting time